Objective This study aims to determine the clinical utility of visual ratings and volumetric measurements of medial temporal atrophy among subjects from the Alzheimer's Disease Neurorimaging Initiative (ADNI) cohort. Methods A sample of 189 subjects from the ADNI, Phase 1 (ADNI‐1), was chosen as follows: 49 cognitively normal (CN), 89 with mild cognitive impairment (MCI), and 50 with Alzheimer's disease (AD). Structural MRI images were downloaded from the ADNI website, and a visual rating system (VRS) was used to obtain semi‐quantitative ratings of the hippocampus (HPC) and entorhinal cortex (ERC). VRS ratings and FreeSurfer measures of the HPC and ERC were used to predict (i) baseline diagnosis and (ii) progression to AD among subjects with MCI at baseline. Results VRS and FreeSurfer measures of ERC were equivalent in classifying subjects at baseline, but FreeSurfer measures of HPC were superior to VRS measures for classifying CN versus MCI subjects. VRS and FreeSurfer measures of both HPC and ERC were significant predictors of progression from MCI to AD. However, VRS ratings of ERC were superior to other MRI measures. MCI subjects with minimal ERC atrophy by VRS had a threefold lower progression rate to AD at 3.2 years compared with those with mild, moderate, or severe atrophy (23% vs 63%, 69%, and 87%, respectively). Conclusions Visual ratings of HPC and ERC provide useful information to a physician in a clinical setting. Visual ratings of ERC may be especially useful in following patients with MCI. Copyright © 2014 John Wiley & Sons, Ltd.